The "eyewitness testimony" that is regarded with justifiable skepticism by forensic psychologists is very different from what a historian might mean when s/he speaks of firsthand and primary sources.

In forensics, there's an experimentally demonstrable problem with eyewitness recall of fleeting events-- man runs into a bank, yells something, a struggle ensues, someone gets shot: bona fide eyewitnesses will predictably have very different stories about what happened. They need not be lying-- human perception of fleeting events isn't a video recording.

Historians don't have much of the legal sort of "eyewitness testimony", because the historian isn't particularly fixated on moment to moment events. We do care a lot about "primary sources", and what people thought and said at the time.

So in interviewing, say, Andrew Goodpaster (one of Eisenhower's aides) -- he had a very clear idea of what the President thought, took notes, prepared position papers and so on. His characterization of "this is how the Administration decided to X”; that's not likely to be mistaken in the way that a eyewitness at a lineup can be mistaken. And of course when we have Eisenhower's own words about why he did what he did -- he may be being disingenuous, but it's certainly fair to say that he didn't have a problem understanding his own mind, even if he may not have always wanted to speak it plainly. Similarly in examining accounts of witch trials, we don't use them as evidence for the existence of witchcraft-- we use them as evidence for the existence of witch _trials_.

One interesting example of an historical episode with an abundance of firsthand testimony whose reliability is in question would be "The Return of Martin Guerre" -- a 16th century case of a possible impostor returning home from war. There's a trial, lots of testimony from people who knew "the real Martin Guerre". There was a book by Princeton historian Natalie Zemon Davis, criticism by other historians like Robert Finlay, and ultimately a movie. As a neatly encapsulated example of the problems historians have in dealing with firsthand testimony, this would be a useful story to examine.

In any case, the firsthand evidence we're looking at in history are generally nothing like a person who pulls into a gas station, sees someone running and is later asked to identify them in a lineup. I'm trying to think of a historical work which relies on that kind of fleeting "eyewitness" -- you get that kind of issue in combat, for example when people try to decide "who shot down the Red Baron", but that's not particularly historically significant.

​

See:

Natalie Z. Davis "The Return of Martin Guerre" (Harvard University Press:1984)

Finlay, Robert (June 1988). "The Refashioning of Martin Guerre". The American Historical Review.